Researchers are often inclined to test agreement or discrepancy hypotheses using difference scores. This commentary explains 2 mathematical-statistical principles underlying associations with difference scores and 2 conceptual-interpretation problems that make difference scores inappropriate for testing such hypotheses. The commentary provides examples of valid and invalid interpretations of difference score associations in reference to equivalent models. The commentary recommends testing agreement hypotheses using interaction terms and explains how to do so. [For Campione-Barr, Lindell, and Giron's paper, "Developmental Changes in Parental Authority Legitimacy and Over-Time Associations with Adjustment: Differences in Parent, First-Born, and Second-Born Perspectives," see EJ1249846.]